Skip to main content

The Experimental AI Microkernel OS

Project description

FyodorOS

PyPI version Python 3.8+ License: MIT

███████╗██╗   ██╗ ██████╗ ██████╗  ██████╗ ██████╗
██╔════╝╚██╗ ██╔╝██╔═══██╗██╔══██╗██╔═══██╗██╔══██╗
█████╗   ╚████╔╝ ██║   ██║██║  ██║██║   ██║██████╔╝
██╔══╝    ╚██╔╝  ██║   ██║██║  ██║██║   ██║██╔══██╗
██║        ██║   ╚██████╔╝██████╔╝╚██████╔╝██║  ██║
╚═╝        ╚═╝    ╚═════╝ ╚═════╝  ╚═════╝ ╚═╝  ╚═╝
          The Experimental AI Microkernel

FyodorOS is a simulated operating system designed from the ground up for Autonomous AI Agents. Unlike traditional OSs designed for humans (GUI/CLI) or servers (API), FyodorOS exposes the entire system state as a Document Object Model (DOM), allowing Agents to "perceive" and interact with the kernel natively.

🚀 Vision

We believe that for AI Agents to be truly useful and safe, they need an environment built for them. FyodorOS provides:

  • Structured Observation: The OS state (Filesystem, Processes, Users) is a queryable DOM tree.
  • Cognitive Loop: Built-in ReAct (Reasoning + Acting) loop at the kernel level.
  • Safety Sandbox: A strict, rule-based verification layer that constraints Agent actions before execution.
  • Agent-Native Apps: Standard tools (browser, explorer, calc) that return structured JSON/DOM instead of plain text, minimizing token usage and parsing errors.

✨ Key Features

🧠 Kernel-Level Agent

The OS integrates an LLM-powered agent directly into the shell.

  • Command: agent "Research the latest news on AI"
  • Mechanism: The agent perceives the system via SystemDOM, creates a To-Do list, and executes actions in a sandboxed loop.

🌐 Agent Browser (Playwright Integration)

FyodorOS includes a specialized browser for agents.

  • DOM Tree Output: Returns a simplified, semantic JSON representation of web pages.
  • Interaction: Agents can click and type using element IDs directly.
  • Efficiency: Strips unnecessary noise (CSS/Scripts) to save context window.

🛡️ Safety Sandbox

Every action taken by the Agent is intercepted by the AgentSandbox.

  • System Protection: Prevents writes to /kernel, /bin, and /etc.
  • App Whitelisting: Only authorized "Agent Apps" can be executed.
  • Transparency: All actions are logged and verifiable.

📦 Installation & Usage

  1. Clone the Repository

    git clone https://github.com/Kiy-K/FyodorOS.git
    cd fyodoros
    
  2. Install Package You can install FyodorOS as a Python package.

    Via pip (Recommended):

    pip install .
    playwright install chromium
    

    Via Conda:

    conda env create -f environment.yml
    conda activate fyodoros
    playwright install chromium
    
  3. Launch the OS

    Option A: Using the CLI (if installed)

    # 1. Setup (Configure API Keys)
    fyodor setup
    
    # 2. Start (Auto-login as Guest)
    fyodor start
    
    # 3. Interactive Login
    fyodor login
    
    # 4. Login as specific user (e.g. root)
    fyodor login --user root
    

    Option B: Using Convenience Scripts

    • Windows: Double-click run.bat
    • Linux/Mac: Run ./run.sh
  4. Interact Inside the FyodorOS Shell:

    # Run a standard command
    guest@fyodoros:/> ls
    
    # Task the Agent
    guest@fyodoros:/> agent "Create a file named hello.txt in my home folder"
    

💡 Use Cases

1. Web Research & Summary

Scenario: You want the agent to look up information and save it. Command:

agent "Go to https://example.com, read the main heading, and save it to /home/guest/summary.txt"

Agent Process:

  1. Calls run_process("browser", ["navigate", "https://example.com"]).
  2. Parses the returned DOM tree to find the <h1> tag.
  3. Calls write_file("/home/guest/summary.txt", ...) to save the data.

2. System Management

Scenario: You want to add a new user securely. Command:

agent "Create a new user named 'developer' with password 'secure123'"

Agent Process:

  1. Checks if user exists using run_process("user", ["list"]).
  2. Calls run_process("user", ["add", "developer", "secure123"]).
  3. Verifies the addition.

3. File Organization

Scenario: Organize a messy directory. Command:

agent "Move all .txt files from /home/guest to /home/guest/documents"

Agent Process:

  1. Calls list_dir("/home/guest").
  2. Identifies .txt files.
  3. Calls run_process("explorer", ["move", ...]) for each file.

🏗️ Architecture

  • src/fyodoros/kernel/: Core logic including Scheduler, SyscallHandler, and the new Agent Layer (agent.py, dom.py, sandbox.py).
  • src/fyodoros/bin/: User-space applications. These are "Agent-Aware" binaries that output JSON.
  • src/fyodoros/shell/: The interactive CLI wrapper.
  • src/fyodoros/cli.py: The launcher and configuration tool.

🤝 Contributing

FyodorOS is an experimental sandbox. We welcome contributions to:

  • Expand the standard library of Agent Apps.
  • Improve the DOM representation of system state.
  • Implement more complex Sandbox rules.

Built for the future of Autonomous Computing.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

fyodoros-0.1.1.tar.gz (24.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

fyodoros-0.1.1-py3-none-any.whl (29.2 kB view details)

Uploaded Python 3

File details

Details for the file fyodoros-0.1.1.tar.gz.

File metadata

  • Download URL: fyodoros-0.1.1.tar.gz
  • Upload date:
  • Size: 24.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.1

File hashes

Hashes for fyodoros-0.1.1.tar.gz
Algorithm Hash digest
SHA256 e8ffe69d3b196cc63ef5cd1edf709d1af24f1e7a6c84bd56e2b85e5ba4fb670e
MD5 744985a60ef4ebd5262d01dabef21d4f
BLAKE2b-256 76489e5b33bc73304e48c65740fcc7157563b0155f3f64269b97bc6b4e3b9f6e

See more details on using hashes here.

File details

Details for the file fyodoros-0.1.1-py3-none-any.whl.

File metadata

  • Download URL: fyodoros-0.1.1-py3-none-any.whl
  • Upload date:
  • Size: 29.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.1

File hashes

Hashes for fyodoros-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 ddeb9424879d0b95194b72545b040837eec0ed9533bd5198f58f5c1b94009f50
MD5 aaef926a9ac616297c50241230234c4a
BLAKE2b-256 86456d56d70b474b5f831d426fc48fd95f441239bd57c8fb2b79d203f101f865

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page